Food Or Fluid Restriction in Common Laboratory Animals: Balancing Welfare Considerations with Scientific Inquiry

Home , Laboratory rat

Overviews Food or Fluid Restriction in Common Laboratory Animals: Balancing Welfare Considerations with Scientific Inquiry

Neil E Rowland

Deprivation or restricted access to either food or fluids is a common research procedure in laboratory animals. The purpose of the present review is to present and summarize some of the important physiologic effects of such procedures and to assess their effect on the well-being of the animal. This assessment is presented within a context of the typical research objectives of such procedures. Specific suggestions are made that are intended to strike a balance between meeting these research objectives and ensuring the physiologic and behavioral welfare of the animals under study. Most of the information presented is specifically related to rats and mice but, with appropriate adjustments, the principles likely will generalize to other laboratory species. I present evidence that after 12 to 24 h without access, animals efficiently reduce further fluid or energy losses by a combination of behavioral and physiologic adjustments. These adjustments likely minimize the additional physiologic or psychologic stress of deprivation. Animals have endogenous nycthemeral rhythms that make them particularly adaptable to once-daily occurrences, such as food or water access. Longer periods of acute deprivation or chronic restriction are acceptable procedures, but only with suitable monitoring protocols, such as routine weighing and target weights. In the case of chronic food restriction, the use of species-, age-, and strain-specific target growth rates is more appropriate than using a fraction of age-matched free-fed animal weights as a target.

Abbreviations: ACUC, animal care and use committee; BMI, body mass index; ECF, extracellular fluid; ICF, intracellular fluid

Animal care and use committees (ACUCs), often working with Normal Feeding and Drinking institutional veterinarians, are charged with ensuring the justified Ecologic considerations. The topic of this paper is restriction use and humane treatment of experimental animals. Some com- or deprivation, terms that can be defined only relative to what monly encountered standards of care and treatment are uncon- is considered ‘normal.’ Often, this state is implicitly assumed to troversial and are relatively easy both to monitor and implement. be unrestricted or ad libitum access to food or fluid but, as will In contrast, many experimental procedures are highly specialized be discussed later, in most cases unrestricted availability is not and often are not within the realm of experience or expertise of optimal for long-term health.26 Continual access certainly is not either the ACUC members or the veterinary staff. The experience how animals would encounter food or water in the real world. and judgment of the investigator are factors that might be consid- The physiologies underlying need differ greatly for eating and ered, but ACUCs are far from uniform in using this information drinking, so in most instances in this review, they will be treated as a basis for their deliberations. To avert potential conflict with separately. investigators, many ACUCs are developing local guidelines for a In the case of eating, most animals have evolved in environ- wider array of procedures. Food and fluid restriction are proce- ments in which availability of food was uncertain and often insuf- dures that are essential to the conduct of certain types of physi- ficient. The body fat content of otherwise healthy animals killed ologic and behavioral research. in the wild vary from 1% to 25%, depending on species and time The purpose of this article is first to present a perspective on of year, and this range is considerably lower than that found in what constitutes ‘normal’ food and fluid intake, then to discuss the same species in laboratory animal facilities.45 Ad libitum feed- the physiologic and behavioral effects of restriction, and finally to ing regimens of nutritious food, standard in most rodent facilities, present some recommendations for standard practice. This treat- does in fact lead to excessive fat deposition and obesity.26,27 Ad ment will be heavily biased toward rodents because these are the libitum food is probably provided mainly as a convenience for most common species used in research. Most of the principles husbandry staff in facilities with large numbers of rodents. In con- likely will generalize to other species with appropriate modifica- trast, many species, including pets, zoo specimens, and livestock, tion including, for example, adjustments based on relative body often are fed only at discrete times. In those circumstances, this size and differences in natural habitat and diet. form of restriction is considered good husbandry. Therefore, the standard is not consistent: ad libitum food is considered normal Received: 20 Jun 2006. Revision requested: 31 Aug 2006. Accepted: 10 Nov 2006. or appropriate for some species but not for others. Although the Department of Psychology, University of Florida, Gainesville, FL. Email: [email protected] convenience of providing food ad libitum for large colonies is 149 Vol 57, No 2 Comparative Medicine April 2007

drink 20 or more bouts per day, most of which are associated in time with their grazing meals (when food has no cost), an environmental procurement cost on water can turn them into once- or twice-daily drinkers. Returning to food, other factors such as type of diet (for example, liquid versus solid), palatability, and ease of access can affect meal patterns in rats.11,68 Mice seem to be even more sensitive in this regard (Table 1). The number of daily meals ranges from 2 to 50, depending on the type of diet and the ease or configuration of its procurement. Even when food is available at all times, mice, like rats, choose to be grazers or gorgers as a function of what to us may seem like modest differences in the environment. Meal pattern depends mainly on factors extrinsic to the animal and affects the detection of experimental deprivation by the animal. A rodent in a grazing mode may physiologically detect the ab- Figure 1. Elective change in mean number and size of daily meal in rats sence of food after only a few hours, whereas a rodent in a gorg- with ad libitum access to food but with different imposed procurement ing mode and eating only 2 meals per day may not detect absence costs (fixed-ratio [FR] lever presses) for each meal. Figure is redrawn of food for 12 h or more. from data in reference 11. Hamsters differ from rats or mice in that they have extremely limited flexibility in their meal size and do not compensate for indisputable, the use of this as a ‘gold standard’ against which to infrequent availability of food by increasing meal size.55 Instead, measure food restriction or deprivation is questionable. hamsters are particularly prodigious hoarders and in natural bur- Meal or bout patterns of ad libitum food or water access. Ro- rows may eat frequently from hoards.4 Therefore, the environ- dents are sometimes referred to as ‘grazers,’ meaning that they mental determinants of meal patterns discussed previously are normally eat frequent but small meals. This term describes the modulated by species-typical constraints. behavior of rodents under conditions of ad libitum and free ac- Endogenous rhythms. Animals have a number of endogenous cess to a diet of relatively high nutritional content, and is not a rhythms ranging in duration from less than 1 d to a year. The most general or intrinsic attribute of rodents, or indeed many other ubiquitous and relevant of these for animal care are ‘nycthemeral species. The actual situation is that most species have flexible eat- rhythms,’ also known as day-night or diurnal rhythms. These ing patterns, adopting different feeding strategies in different rhythms are temporal organizing mechanisms for physiology and environments. behavior in all species. Most rodents are nocturnally active, so the Under typical laboratory conditions of free access to a nutri- majority of their food foraging and intake occurs at this time. For tionally balanced food such as a commercial chow, which yields example, in the reports of mouse meal patterns summarized in 3 to 3.5 kcal metabolizable energy per g,46 rats adopt a grazing Table 1, most reported that approximately 75% of the meals were pattern in which they eat about 10 meals daily. Most laboratory taken during the lights-off phase of a 12:12-h cycle. In rats, 6 to rodents are nocturnal, so most of these meals normally occur at 9 h may elapse between the last meal taken near the time of light night (see also the following section, Endogenous rhythms). The onset and the next meal; conversely, an intermeal interval of more average meal size is 2 to 3 g, and so the total daily intake is 20 to than 2 h is unusual during the lights-off phase.30 30 g (approximately 60 to 90 kcal). Figure 1 shows that rats can Nycthemeral rhythms are maintained by internal oscillators easily be changed from this pattern to a ‘gorging’ strategy. In the whose periodicity is approximately 24 h, and they normally are experiment shown,11 rats lived in an environment in which they entrained to the light-dark cycle in a species-typical manner.35 could work or forage for food (complete rodent diet) at any time However, animals can readily change the timing or phase of their they desired. The independent variable was ‘procurement cost’: feeding to other times of the day-night cycle, for example if food the rats were required to press a lever a fixed number of times to is available only during the daylight hours or at a fixed time each open a door that gave them access to a food dish. They could eat day. The phenomenon of jet lag experienced by many travelers as much as they wished from the dish for no additional cost but, is due to the fact that the internal clock normally can only be ad- once they left the feeder for 10 min, the door closed. In order to justed by 10% (2 to 3 h) each 24-h cycle.35 Likewise, nocturnal eat again, they would have to ‘pay’ another consumatory cost. rodents require several days to adapt to either restriction of food Rats were studied for several days at each of a range of costs. As or feeding during the daytime; such adaptation may be impaired procurement cost increased, the mean number of meals per day in conditions of brain damage or drug action.54 During restriction decreased (Figure 1, left panel) and their size increased (right pan- paradigms, animals are usually fed during the daylight hours el), so that daily total intake was approximately constant. These for experimenter convenience. Using reversed light-dark cycles are all ad libitum feeding patterns because food was accessible at for feeding studies matches the time of food availability to the any and all times. The grazers became gorgers because of a small natural rhythm of feeding, although not all facilities may be able access cost, probably a modest 10 min or less of lever pressing to accommodate this provision. daily. If we establish an access cost that causes an animal to eat Physiologic cycles of metabolism underlie the spontaneous voluntarily only 1 large meal per 24 h, it is difficult to argue that day-night feeding patterns of rats.30 At night, rats eat more food this is different physiologically from a feeding schedule in which than the energy they expend, and so the excess is stored as gly- an investigator chooses to feed an animal once daily. cogen and fat (lipogenesis). By day, rats eat less food than the Similar data have been presented for procurement costs and energy they expend (mostly basal metabolism, because they are water intake.33 That is, although under ad libitum conditions rats resting), and so they mobilize glycogen and fat (lipolysis) reserves 150 Food and fluid restriction

Table 1. Ad libitum meal patterns in mice

Mouse strain Feeding configuration Diet No. of meals/24 h (end criterion) Reference Small (S) and large Overhead door panel Powdered rodent 12 (5 min) 44 (L), inbred chow SWR/J Recess at floor level Powdered rodent 36 (5 min) 18 chow C57BL/6J (lean and Sipper spout in cage Liquid diet EC116 50 ( ) 30 () (various) 59 ob/ob) C57BL/6J (lean and Lever press and food Noyes 20-mg pellets 2–10, function of access cost (10 min) 65 ob/ob) receptacle 129/B6 (wild type Pellet removal from trough BioServ 20-mg _12 (18 h food access) 16 for BNDF /–) pellets 129/B6 (wild type Liquid diet from 0.02-ml Isocal high-fat liquid _15 (18 h access) 16 for BNDF /–) dipper 129/B6 (wild type Lever press and food Noyes 20 mg pellets 2–7, function of procurement cost (10 min) 63 for MC4R–/–) receptacle (procurement cost) 129/B6 (wild type Lever press and food Noyes 20 mg pellets 25–50, function of reset time (20 versus 3 min) 64 for MC4R–/–) receptacle (progressive ratio) via sympathetic neural and hormonal mechanisms.30 As Fried- This engineering concept translates reasonably well to biologic man and Stricker17 so vividly state “the animal with no food to systems that have little or no capacity for storage; such systems consume must consume itself.” These cycles of consuming from include respiration, fluid homeostasis, and thermoregulation. the outside and consuming from the inside are in fact natural Feeding or energy balance is a quite different matter, because of cycles (24 h and shorter), the periodicity and amplitude of which the enormous capacity for storage and mobilization, as mentioned depend on environmental factors (for example, as in grazing ver- previously. For a given level of food or energy intake, energy ex- sus gorging). penditure will stabilize at a particular level, and this level will be Like food intake, spontaneous or ad libitum water intake in rats associated with a particular size of the energy stores. Figure 2 is a is mainly nocturnal, most of it associated with meals.28 Spontane- theoretical analysis of a situation that does not include a set point ous water intake can to some extent be dissociated from avail- for stores or body weight22 and shows that in principle an infinite ability of food: when food intake of rats was equally rationed number of steady states are possible. Therefore the concept of set between day and night, water intake was still 75% nocturnal.37 or settling point for body weight20,38,69 has little or no meaning However, the timing of food intake when water is restricted may without specifying the environment to which it refers. Relevant be less determined by the day-night cycle, in part because dehy- environmental factors include but are not limited to the nutritive dration causes anorexia (see the section titled Water deprivation value or composition of food, its palatability, the relative effort and body fluids). expended in food procurement, ambient temperature, and avail- Some laboratory animals (for example, hamsters, squirrels, ability of water. some avian species) show circannual rhythms, even in a labora- A demonstration of the flexibility of body weight is the para- tory with constant temperature and day-night cycle. One feature digm of dietary obesity in which rodents fed a highly palatable, of these species is that they gain weight at 1 phase of the year and calorically dense diet eat more calories and gain body weight and lose weight at another, and studies using food restriction may fat compared with chow-fed controls.62 Conversely, poor tasting need to take this natural variation into account. or calorically dilute diets may be consumed in smaller amounts Another cycle is that related to the hypothalamo-pituitary- and produce lower body weights.43 gonadal axis in female animals. In rats and mice, estrous cycles Rodents and other species that are maintained on a monoto- normally are 4 d in duration, characterized by increased plasma nous or bland diet such as chow will avidly consume a palatable concentrations of estradiol and progesterone, increased locomo- treat that is presented for a short time each day. Such treats can be tor activity, and decreased food and water intake. Most of these used as experimental meals57 or as environmental enrichment.51 effects are due directly to estradiol.13 Therefore, female mice In the latter study, mice compensated for the calories in the treat and rats will have natural cycles of 4-d food intake and body by reducing their ad libitum intake of chow by the same number weight, and studies using food deprivation or restriction may of calories. However, if the treat provides high caloric yield, either need to account for this variation by prior monitoring to deter- because of its high fat content or an extended period of avail- mine or control the phase of the cycle at which experiments will ability, then fully compensatory caloric reduction in chow intake be conducted. may not occur. Treats presented either as meals or as enrichment Set points. The mean values around which body weight oscilla- generally are not intended to produce dietary obesity. tions occur often have been called ‘set points’ by analogy with the engineering systems that are designed to regulate variables such Deprivation and Restriction as temperature via thermostatic control system in an animal facil- Toth and Gardiner61 drew an important distinction between ity. A person sets the desired temperature, and the system is de- deprivation and restriction, and I will discuss that general distinc- signed to deliver an acceptable oscillation around that set point. tion as well. 151 Vol 57, No 2 Comparative Medicine April 2007

Table 2. Body fluid compartments and constituents in humans

Physiologic property Intracellular Extracellulara Volume (% body weight) _40% _20% Na (mEq/l) 12 145 K (mEq) 150 4 Ca (mEq/l) 0.001 5 Cl– (mEq/l) 5 105

Phosphates (Pi, meq/l) 100 2 aThe extracellular fluid compartment is_ 75% interstitial fluid and_ 25% blood plasma. Values for most mammals, and in particular rats and mice, are quite similar.

of the risk of fighting and/or because unequal competition for a ration offered to a group of animals will introduce uncontrolled variance in the intake of the individuals in that group. Figure 2. Optimal food reserves (mostly as triacylglycerols in adipose The use of a modest amount of a palatable and nutritious food tissue) as a function of relative energy gain from foraging in a theoreti- treat51 may be incorporated into some restriction studies. The use cal model of energy flow into and out of a virtual animal comparable in of chewable objects may also be acceptable, although whether size and physiology to a rat. The relative energy gain is related to the their provision affects food motivation has not been assessed for- energy in the food minus the energy expended to obtain that food. Note mally. The use of larger cages or activity devices may affect ener- that expected stable body weight is a continuous function of net energy gain; no body weight ideal or set point is built into the model. Figure is gy expenditure during food deprivation, but this possibility also redrawn from data in reference 23. has not been evaluated comprehensively. In-cage shelters such as polyvinyl chloride pipes or other plastic enclosures are well used 19 Deprivation studies. Many protocols are designed to evaluate by rodents, but whether this use affects energy loss during food restriction, for example by reducing local heat loss to the room, the behavioral or physiologic effect of withholding a commodity 51 for various periods of time. The overwhelming majority of stud- has not been examined. My laboratory examined weight gain ies in the literature has used multiples or submultiples of 1 d (for in mice with or without plastic ‘igloo’ shelters and on 2 diets and example, 6, 12, 24, 48 h). These are known as ‘deprivation studies’ found no significant effects of the shelters. However, the general- because at the end of the designated period, the animals are test- ity of that finding must be established in a wider range of condi- ed and returned to free access of the commodity or, in some cases, tions before it can be recommended for routine or general use. are euthanized. Animals may experience some discomfort during longer deprivations, but the deprivation period has a defined end Physiology of Water Deprivation and Restriction time. For this type of study, the issue before the ACUC is the sci- Body ﬂuid compartments. Body fluids are distributed both in- entific justification for a particular duration of deprivation within side of cells (intracellular fluid, ICF) and outside of cells (extra- the context of potential distress or physiologic harm. cellular fluid, ECF). The relative sizes and ionic compositions of Restriction studies. Restriction studies, in contrast, do not in- these 2 compartments are shown in Table 2. Fluids are lost con- volve complete withholding of a commodity but rather the tinuously in urine and feces, as sweat, and by respiration.7,15,56 presentation of a controlled ration each day for a more or less pro- Water and solutes move between ICF and ECF under the force longed period of access. Most studies of this type use the restric- of osmotic pressure. Net movement does not occur when the os- tion protocol to reproduce a consistent state of physiologic need motic pressure is equal on both sides of cell membranes. Osmotic from day to day such that behavior will be motivated and stable. pressure is normally about 290 mOsm/l, which is often known This state is a prerequisite for scientific evaluation of hedonics, as ‘isotonic.’ ECF has both interstitial and plasma components. reinforcement, and other higher cognitive and motivational char- These are similar in composition because they are connected acteristics in animals. The chronic nature of these studies could through the relatively porous junctions between endothelial cells have adverse effects over time; the health status of such animals that comprise capillary walls, but molecules such as plasma pro- should be monitored, for example by weighing or physical exam, teins (for example, albumin) are trapped in the vasculature be- and observations or measurements recorded at intervals deemed cause they are too large to leak out. The concentration of plasma appropriate by the ACUC. Restriction studies normally are per- protein rises if ECF is lost from the vasculature, thus providing a formed using healthy animals in which the physiologic conse- useful indicator of intravascular dehydration. quences differ from those of anorexia due to illness. A healthy ECF is filtered into tubules in the kidney; the amount of urine animal that has lost 15% body weight by restriction is likely to be formed at the other end of the nephron is critically dependent on in a clinically stable condition, whereas one that has lost the same 2 main types of reabsorptive processes that occur in the tubule. weight due to illness is not. First, sodium (along with chloride and water) is reabsorbed in Environmental enrichment considerations. The Guide for the Care the proximal tubule; this process is stimulated by aldosterone. and Use of Laboratory Animals40 recommends that social housing is Second, water is recovered in the distal tubule by the action of the best form of environmental enrichment for social species, in- vasopressin (antidiuretic hormone) at V2 receptors, thus concen- cluding rodents. Social or other enrichment can affect the impact trating the solutes in the remaining tubular fluid, which becomes of restriction or deprivation in several ways. For example, social urine. The maximal concentration of urine that can be achieved housing is not recommended for animals on restriction because by rodents is approximately 3000 mOsm/kg water, about 10× 152 Food and fluid restriction

Table 3. Effects of various durations of water deprivation on activity, food intake, body weight loss and recovery in Wistar rats Duration of water deprivation Home cage Food intake Weight loss Days to weight recovery (h) activity (%) (% baseline) (% baseline) (99%) 0 (baseline days) 100 100 0 0 24 96 88 4 1 48 104 28 8 3 72 99 14 11 5 96 65 14 13 9 Values are derived from graphical data presented in reference 1. Home cage activity and food intake are changes across time in a single group deprived for 96 h. Weight loss and recovery data (to 99% of growth-extrapolated baseline of ad libitum group) are from 4 separate groups differing in duration of deprivation. isotonic, but may decline with age.6 The physiologic impact of in Figure 3. Urinary volume was reduced by more than 50% dur- water deprivation will depend critically upon the magnitude or ing the first 24 h, yet urinary concentration (osmolality) was only efficacy of these 2 mechanisms of fluid conservation. These pro- slightly increased. This difference is due to the large reduction in cesses are extremely efficient in many species, although genetic food intake and to a urinary output of Na and K that exceeded or other experimental manipulations have the potential to reduce the intake in food. Others have reported that the natriuretic effect this efficiency, and such animals should be monitored particularly of water deprivation, amounting to 3 to 4 mmol/kg after 24 h in carefully. rats, is similar in rabbits, sheep, and dogs,32,34 and that it occurs Water deprivation and body ﬂuids. The physiologic effects of using a variety of diets.33 This pattern is a direct effect of dehy- water deprivation are critically dependent on composition of dration and not the associated anorexia because comparable na- the maintenance diet and whether it is available during the de- triuresis was not observed when hydrated rats were pair-fed to a privation period. Most rodents are maintained on a commercial water-deprived group.32 When rats were water-deprived without chow formula. Purina Mills Inc. reports that their #5001 diet con- food available, the physiologic changes were qualitatively similar tains (by weight) 0.40% Na , 0.95% Ca , 1.10% K , 0.65% Cl–, to but smaller than those in rats deprived with food available.5 and 0.67% P.46 Diets like this have a fairly high fiber content and This urinary ‘solute dumping’ along with reduced food intake produce high fecal volume and corresponding fecal water loss and fluid loss minimize the effect of water deprivation on body (approximately 2 g/g fecal solid), amounting to approximately fluid homeostasis. During water deprivation in rats with food 10 ml/d in an adult rat, or 25% of their total daily water loss.47 present, the increases in plasma osmolality were only a modest With a low-fiber synthetic diet, fecal volume and water loss is approximately 2% and 4% after 24 and 48 h, respectively;5,53 the much lower. corresponding decreases in plasma volume were approximately In addition to the hormonal conservation responses mentioned 4% and 13%, respectively. Changes in parameters of hydration af- in the section on body fluid compartments, a critically important ter 24-h water deprivation in dogs, monkeys, rabbits, sheep32,53 are behavioral response that rodents make during water deprivation similar to the cited rat data and may be less than in humans.61 is to reduce food intake. This compensatory action is referred to Mice have considerably smaller body size and correspond- as ‘dehydration anorexia.’66 This action serves 2 purposes: first, it ingly faster water turnover, compared with rats. In a survey of reduces the quantity of electrolytes that enter the body and ulti- 28 strains of mice fed chow,2 the mean daily water intake varied mately would have to be excreted, and second, it reduces obliga- by 2-fold across strains but averaged 7.7 ml/30 g body weight, or tory fecal volume and the associated water loss. This anorexia about 25% of their body weight. In contrast, rats drink approxi- becomes progressively greater with increasing duration of depri- mately 10% of their body weight daily.28 Mice also show larger vation. For example, Bealer and colleagues5 reported that chow changes with dehydration. Therefore, 24-h water deprivation of intake of male rats fell to approximately 65% and 30% of the ad various wild-type or control strains of mice with dry food avail- libitum level during the first and second 24 h of water depriva- able produces weight losses of 6.7% to 9.5%21,50,52 and a 6.2% in- tion, respectively. Comparable data were reported by Armstrong crease in plasma osmolality.52 Both of these changes are about and colleagues,1 who extended the deprivation to 96 h (Table 3). 2-fold greater than those summarized previously for 24-h water- They found that food intake was even lower on the 3rd and 4th deprived rats.5 The relative anorexia during water deprivation days of water deprivation but that body weight loss slowed from in mice50 was comparable to that reported for rats. Mice with a 4% on each of the first 2 days to 3% and 2% on days 3 and 4.1 Loss genetically engineered change in a component of fluid balance of body weight therefore is not an accurate index of water loss may show impairments in fluid conservation or other aspect of or dehydration, because most of the loss is the result of anorexia fluid homeostasis, and so studies that include such strains should and decreased food volume in the gastrointestinal tract. Weight consider carefully the additional physiologic effect of a given loss may be more useful as an index of nutritional status. In rats, duration of water deprivation.36 Because 24 h of water depriva- the weight loss observed after 96 h of water deprivation (13%) is tion has apparently larger physiologic effects in mice than rats, approximately the same as that after 48 h of food deprivation,1 investigators planning to extrapolate from rat to mouse studies suggesting that water deprivation is physiologically less stressful should be aware of this difference. In the absence of comparative than food deprivation. Activity level in that study did not de- time-course data on the physiologic effects of water deprivation crease until the 4th day of water deprivation (Table 3), further in mice versus rats, weight loss may be a suitable indirect crite- suggesting that these animals remained physically capable for at rion or target. least several days. A detailed time-course of changes during deprivation has been The urinary exchanges reported by Bealer and colleagues5 dur- reported for dogs, in which repeated blood sampling is feasible ing 48-h water deprivation in rats with chow available are shown without undue impact on hydromineral balance. Reviewing these 153 Vol 57, No 2 Comparative Medicine April 2007

from the data presented earlier, causes depletions of both ICF and ECF. The thirst and eventual water intake observed is the sum of the signal from these 2 components.15,48,53,56 Water restriction. Water restriction can be useful as a motivat- ing stimulus in behavioral studies.21 Most animals adapt well to once-daily access to water, and this frequency seems to be common for primates in arid natural habitats.12 Hughes and col- leagues23 reported intakes and weight changes of rats during 2 mo of water restriction in a water-rewarded operant task. Groups of rats were deprived for 7, 14, or 21 h daily. Operant groups received a 1-h test session at the end of this time and then had free water and food in their home cage for the remainder of the scheduled access. Nonoperant groups also were run and simply received water restriction for the same durations. Their weight changes are shown in Table 4. Only the 21-h group showed an initial small weight loss and a modest reduction in weight gain during the study. However, only the 21-h operant group was able to acquire the water-motivated task, indicating that this level of water deprivation has distinct motivational characteristics, com- 9 Figure 3. Relative mean changes in urinary excretion of sodium and pared with 7- or 14-h deprivation. Likewise, Carlton showed that potassium (both as % of the intake of these cations in food available responding for water on interval schedules was a steep function during this period) and volume and osmolality of urine (as % baseline) of deprivation time, with a maximum after 23 h. Although the in rats during 2 consecutive 24-h periods of water deprivation. Figure is timing and amount of food consumed usually is not measured, redrawn from data in reference 5. the weight gain data given by Hughes and colleagues23 indi- cate that dehydration anorexia was minimal. Other studies have data may help inform rodent studies. Parts of these data34 are shown that water-deprived rats drink and then alternate eating redrawn in Figure 4, which shows changes in blood indices of and drinking,67 so most of the food intake likely occurred dur- fluid balance across 24 h of dehydration and then during 24 h of ing the period of water access in the home cage. Kakolewski and rehydration. Not shown are values from a preceding baseline day Deaux25 reported that rats accustomed to a 15-ml ration of water which, for the most part, were steady at or near the initial values once daily ate 4.2 g dry food during the ensuing hour; this quan- for the dehydration day. Figure 4 shows that during water depri- tity is not as much as their expected total daily intake but is a sub- vation, plasma osmolality increased rapidly and peaked at 7% to stantial meal. Therefore, animals seem to be well able to entrain 8% above baseline after approximately 10 h; no further increase their ingestive responses to these times of availability without occurred after that time. Plasma sodium concentration showed a evidence of compromised health.23 Data relating the time course similar trajectory (data not shown), and plasma vasopressin lev- of changes in body fluid parameters during this type of chronic els followed osmolality quite closely. This association is expected restriction have not been published, but as for acute deprivation, because vasopressin secretion is under the direct control of osmo- the type and timing of food access would be key variables. Even receptors.15,56 In contrast, the concentration (or, more accurately, mice, with their small body size and relatively larger effect of the enzymatic activity) of plasma renin, a hormone secreted in re- 24-h water deprivation (as discussed earlier), survive severe wa- sponse to decreased plasma volume, increased more slowly dur- ter restriction schedules,41 although the ration in the cited report ing deprivation. Aldosterone, a hormone of sodium deficiency, should not be used routinely. was unchanged. Therefore, plasma osmolality appears to be the Because many behavioral experiments are performed only 4 earliest indicator of dehydration. However, the relative timing or 5 d per week, some investigators allow animals ad libitum ac- and magnitude of these changes may have been affected by feed- cess to water over the weekend. This practice allows recovery ing. The dogs received a moist ration at time 0 on all days and of food intake, and body weight of these animals shows a ‘scal- consumed it within a few minutes. If the diet had been provided loping’ effect. However, whether this pattern is an improvement at a later time of day, then the rise in these parameters presum- over restriction on 7 d per week is debatable; primates, including ably would have been correspondingly later. humans, living in chronically water-restricted conditions adapt On the rehydration day, the dogs consumed approximately physiologically and behaviorally to the routine, so breaking that 1500 ml (and their food ration) during the first 2 h, and the excess routine may not in fact be beneficial.12,42 water ingested balanced their deficit from the deprivation period. Recommendations regarding water deprivation or restriction. As a result, plasma osmolality and vasopressin returned rapidly 1) Rats and mice are physiologically equipped to tolerate acute to normal (Figure 4). In contrast, renin remained high for more dehydration for periods of as long as 24 h without overt signs than 12 h, and aldosterone showed a large and enduring increase, of physiologic distress or behavioral abnormalities. 2) Water presumably in part to retain sodium lost during deprivation, and deprivations in excess of 24 h produce only small additional with the eventual effect of restoring plasma volume. changes in hydration level but do produce substantial anorex- Acute depletions of either the ICF or ECF evoke thirst. These ia. Therefore, for most purposes, deprivation in excess of 24 h effects can be studied independently, and such experiments show likely is not necessary. Because most food intake occurs at night, that the threshold for ICF or cellular dehydration is approximate- overnight deprivation with food available should produce com- ly 2% whereas that for ECF or extracellular dehydration is typi- parable dehydrating effects to a full 24 h of water deprivation. cally a little higher.15,56 Natural water deprivation, as is evident Deprivation of as long as 72 h is tolerated by rats, with weight 154 Food and fluid restriction

Figure 4. Changes in plasma osmolality and renin activity and concentrations of aldosterone and vasopressin during 24-h periods of water deprivation (dark symbols) and rehydration (light symbols) in dogs. Animals were fed once each day at time 0. Figure is redrawn from data in reference 35. loss in an acceptable range (approximately 11%) and no appar- Physiology of Food Deprivation and Restriction ent loss of physical vigor. However, the need for such longer Energy compartments and expenditure. The section on normal durations should have strong scientific justification. Shorter feeding and drinking introduced some general principles about times may be more suitable for smaller rodents, such as mice, cycles of food intake and energy storage or mobilization. About although data are lacking. 3) Because of endogenous circadian 50% to 75% of the energy expenditure under normal laboratory rhythms of physiology and behavior, deprivation or restricted conditions is basal metabolism, the energy used to maintain body access of 24-h duration and periodicity often is particularly well temperature and resting biologic processes. The thermoneutral accommodated. 4) Animals adapt well to once-daily water re- range is the range of environmental temperature under which striction schedules but should be allowed sufficient access time basal metabolic rate is approximately constant. The Guide for the each day to allow them to consume at least some food while Care and Use of Laboratory Animals40 has recommended ambient water is available and so that established criteria for food intake temperature ranges for the maintenance of laboratory species, or weight change are satisfied. Suitable criteria might be that and these are in part based on published thermoneutral ranges. the animal is eating 75% of ad libitum food intake or loses less Although many animals can survive for long periods outside of than 15% weight, but these values are offered only as examples these ranges, basal metabolism is higher, especially in the cold and may differ between species or with different types of food. when extra body heat is lost to the environment. Most animals in- 5) Water-restricted animals should be weighed regularly and crease their daily food intake substantially in the cold.8 Pregnancy provision made in the protocol for supplemental water if the and especially lactation are other circumstances of increased en- approved criteria are not met. ergy loss, in this case to the offspring, and these conditions are

155 Vol 57, No 2 Comparative Medicine April 2007

Table 4. Effect of various durations of chronic water restriction in in the home cage and increased progressively above baseline, ovariectomized female Long-Evans rats particularly during the daytime. Under ad libitum conditions, Minimum body Body weight (% daytime activity was approximately 25% of total, but by the 4th Restriction weight (% initial) initial) 60 d later day of deprivation, it had risen to almost 50%. None 100 120 Severinsen and Munch measured activity and core temperature 58 7 h/d 100 117 of Wistar rats continuously using a small implanted transmitter and obtained a slightly different result. One group of rats was 14 h/d 98 121 food-deprived for 9 d, whereas another group was fed 25% of 21 h/d 96 114 the ad libitum ration every morning. Over this time period, the Data are derived from reference 23, Figure 3 A. authors reported a 27% loss in body mass in the animals starved for 9 d and a 13% loss in the restricted rats. Core temperature associated with increased food intake. Husbandry programs of- continued to show a circadian rhythmicity, and mean daily tem- ten recommend feeding pregnant rodents a diet of higher energy perature decreased by up to 0.4 pC in the restricted group and 1.1 yield. Physical activity typically accounts for less than 25% of the pC in the starved group. This change in temperature was consis- energy expenditure of sedentary rodents, and this proportion can tent with a previously reported decline in resting metabolic rate be increased by regimens of vigorous exercise. Exercising animals of 22% and 31% after the ninth day of restriction or deprivation.39 do not always increase their food intake, and in this event they In contrast to the result of Armstrong and colleagues1 mentioned 14 maintain leaner body weights than sedentary animals. earlier, locomotor activity in the Severinsen and Munch study de- To a first approximation and over a suitably long time frame (at creased by approximately 20% during the first day of deprivation least 24 h), any mismatch between energy expended and energy and by 34% and 48% of control by the ninth day of restriction or intake results in a reciprocal change in nutrient stores of the body. deprivation, respectively. Locomotor activity during deprivation The primary nutrient store is adipose tissue. During food depriva- is likely to depend on measurement technique and ambient tem- tion, adipose tissue will be the principal source of calories to fuel perature. Severinsen and Munch58 used a relatively high ambient metabolism. A progressive and adaptive decrease in metabolic temperature of 28 pC, which reduces energy loss by thermal con- 30 rate occurs as the duration of deprivation increases. Therefore, ductance that occurs in proportion to the difference between body during chronic food restriction, a new equilibrium is established and ambient temperatures. Therefore, food deprivation at higher between the reduced intake and the reduced expenditures to ambient temperature will result in greater metabolic savings and 29 achieve stable body weight. Across a wide range of conditions, less weight loss than would the same duration of deprivation at 0.75 metabolic rate is proportional to (body weight in kg) . low ambient temperature. Studies proposing prolonged food re- Physiologic aspects of food deprivation. The amount of energy striction or deprivation should specify and monitor the ambient stored in adipose tissue varies extensively between individuals temperature. of a species, between species, and as a function of environmental In these examples, after the deprivation period, animals readily conditions, so the physiologic impact of food deprivation will in refeed and gain back most of the lost weight within a few days part depend on initial adiposity. For example, a 30-g mouse with (Table 3). Even relatively prolonged food deprivation appears to 10% body fat content (3 g) has approximately 27 kcal stored as fat. have no long-term adverse physiologic effects. Nonetheless, in- Mice typically eat 12 to 14 kcal/d; thus this adipose store amounts vestigators should always use the shortest period of deprivation to 2 d of ad libitum food intake. By the same calculation, a geneti- consistent with study objectives. Lastly, as is the case for water de- cally obese mouse weighing 60 g with 55% body fat content (33 privation, food deprivation in multiples of 24 h are scientifically g) has approximately 297 kcal stored as fat; even if these mice ate the most defensible because circadian variables are minimized; twice as much as lean mice (they do not), this quantity would be food deprivation for 24 h leads to less than 10% weight loss in equivalent to more than 11 d of stored energy. For a larger animal, most species, and this amount should routinely be acceptable. a 300-g rat with 10% body fat will have 270 kcal stored, and at Longer periods need careful justification and review, including a 65 kcal/d food intake, this amount is equivalent to 4 d of stores. daily surveillance protocol and end points. These oversimplified examples, which among other things do not Physiologic aspects of food restriction. Food restriction offers a account for reduced metabolic rate during deprivation, show that more complex set of issues than total deprivation. Restriction reg- larger animals either by species or by fatness are better able to imens most often are used for behavioral experiments in which a withstand starvation than are smaller animals. Young and grow- consistent level of motivation is needed from day to day. Animals ing animals have a smaller percentage of body fat than do adults often are food-deprived for 16 to 22 h and then are tested in a and thus are more vulnerable during starvation. It follows that behavioral task in which food or another commodity, such as a any consideration of the physiologic and possibly the psychologic drug, serves as a reinforcer. Either immediately after the session effects of food deprivation must consider the age and initial state or after some delay the animals are fed a free daily ration that or adiposity of the animal. complements the amount consumed during the test session. The 1 Armstrong and colleagues reported the effects of food depri- total amount of food eaten either is held constant or varied slight- vation for as long as 4 d in rats; selected aspects of those data are ly depending on whether the animal is above or below a target shown in Table 5. Weight loss occurred as a negatively accelerated weight.9 Most investigators either have developed or follow pro- function of time, with approximately 7% of initial weight lost tocols that use a food ration that has been documented previously occurring in the first 24 h. Some ACUCs use weight loss criteria to produce stable performance in behavioral tasks. This type of such as 10% or 15%, which correspond to slightly less than 48 study often takes months or even years; therefore a question that and 72 h of food deprivation, respectively, in rats of this age (ap- invariably arises is whether either the proposed level of rationing proximately 5 mo) and initial weight (approximately 440 g). In the might be less severe, or whether rationing is necessary at all, in cited study, activity was measured by using a movement detector order to achieve the scientific objective. 156 Food and fluid restriction

Table 5. Effect of various durations of food deprivation on body weight loss, home cage activity, and recovery in Wistar rats Weight loss Days of refeeding needed to reach 95% of Duration of food deprivation (h) Activity (% baseline) (% initial) initial weight 24 7 106 1 48 12 112 6 72 17 116 11 96 19.5 130 8 Data are derived from reference 1.

First, let’s examine the criteria for maintaining such animals. In adult humans and rodents, changes in weight are reflected ACUCs often use a criterion of weight loss (for example, 10% as a proportional change in body mass index [BMI], so the 74% or 15%). Recall from the earlier discussion that these percentage increase in BMI between 33 and 113 wk in female rats [Table 6], losses correspond to those after 24 to 72 h acute deprivation in translated to human scales, would certainly be alarming to a phy- rats, recognizing that there may be species and strain differences. sician. A 75% ration slows weight gain and allows a reasonable Many animals are started in these experiments in young adult- body fat content. A 50% ration maintains a lean but very healthy hood, but if freely fed from that age, they would probably have rat. Such restricted feeding regimens may not be feasible for use continued to grow for some time. in routine animal husbandry and may not be necessary in the A recent experiment using Sprague-Dawley rats clarified the majority of studies, which are relatively short term, but these relationship between long-term rationing and consequences for data show clearly that using ad libitum-fed animals as a ‘gold body weight.27 Parts of these data are summarized in Table 6. standard’ against which to gauge the effects of food restriction First, in the ad libitum-fed groups, both male and female rats in prolonged studies is not appropriate. The gains shown during gained weight throughout life. In male rats, this gain was more aging in the 75% group (Table 6) seem to offer more reasonable rapid during the first year, whereas in female rats, it occurred standards for growth in rats and may be suitable for mice. prominently during the second year, possibly as a result of losing A final point on this topic is obvious individual variation. For ovarian cyclicity and estrogen, a known appetite suppressant. example, considerable ranges in body weight at 7 wk of age are Second, in both male and female rats, the percentage of body fat shown in the footnote to Table 6. Therefore, expressing target approximately doubled between young and mid-to-late adult- weights as the percentage of a group mean may be inappropri- hood. Third, this gain of weight or fat was associated with high ate: for an animal that begins restriction at 10% above the mean blood cholesterol and triglycerides, organ pathologies, loss of es- weight, a weight restriction to 20% below the mean represents trous cyclicity in more than 50% females by 9 mo of age, and high a restriction to 30% for that animal. Conversely, for an animal mortality (only approximately 18% survived for 113 wk). that starts at 10% below the mean, a weight restriction to 20% The most stringent food restriction (48% of initial ad libitum in- below the mean represents an individual restriction of only 10%. take) was associated with slow weight gain throughout the study. Reliance on mean growth charts may introduce unacceptable in- If this gain was expressed as a percentage of the ad libitum values, dividual variability in actual weight loss (a 10% to 30% range in the restricted rats appeared to lose weight but only because the ad this example) and, most likely, in the resultant motivational state. libitum comparison groups continued to gain weight. In contrast To minimize but probably not eliminate this variability, I advocate to that of the ad libitum-fed rats, the body fat of the 48% groups restriction regimens that target a weight gain trajectory based on remained constant and low throughout the study. These rats also the initial weight of each individual. Sensible targets in rats might had low blood cholesterol and triglycerides, fewer organ patholo- be 2 g/wk in males and 1 g/wk in females. These levels corre- gies, sustained estrous (approximately 25% rats cycling even at 2 spond to the average weight gains between weeks 20 and 113 in y), and low mortality (approximately 80% surviving at 113 wk). the 75% groups from Table 6. Intermediate food restriction (75% of initial ad libitum intake) Psychologic aspects of food restriction. In addition to the physi- produced intermediate effects. The report of Keenan and col- ologic aspects of food restriction, ACUCs often consider the psy- leagues27 actually used 2 groups restricted to about this extent, chologic aspect. This task is extremely difficult and subjective but for simplicity I have discussed only 1. Male rats gained some because in the absence of verbal reports from the animals, humans body weight and fat content during the experiment (mean weight often anthropomorphize. The field of food-motivated behaviors at week 113 was 146% that at week 20), and their body weight in animals gives important insights into their subjective states remained at a constant proportion (78% to 84%) of ad libitum and so may be an objective lens through which we can examine values. Female rats also gained weight (mean weight at week questions about psychologic aspects of food deprivation and re- 113 was 136% that at week 20) to a level comparable to that of striction. For example, although humans can easily discriminate males. However, unlike males, female rats maintained body fat between 2 h and 22 h of food deprivation, painstaking training expressed as percentage of weight, but their weight as percentage is necessary to get rats to reach a criterion of only 80% correct.24 of ad libitum values fell due to gain by the ad libitum groups. Therefore, asking animals “How hungry do you feel?” in this way Keenan and colleagues26 have advocated the use of partial re- will not produce a quick or infallible answer. striction to 75% of ad libitum intake in long-term studies that re- Performance on interval schedules of reinforcement is quite quire maintaining a healthy population for years (including, for sensitive to deprivation level. Clark10 maintained rats at slightly example, toxicologic or behavioral work). Their recent results27 less than 85% of free feeding by feeding them once daily with a show that ad libitum feeding of chow, a relatively low-fat diet, calculated ration of dog chow. The rats first were trained by using to rats over a lifetime produces body fat levels that in humans variable-interval schedules (1, 2, or 3 min) of food reinforcement. would be considered obese with attendant metabolic problems. In this type of schedule, food is delivered contingent on a lever

157 Vol 57, No 2 Comparative Medicine April 2007

Table 6. Effect of chronic dietary restriction on mean body weight (g and % of ad libitum) and body fat (% body weight) in Sprague-Dawley rats Male rats Female rats Chronologic age _75%, _48%, _75%, _48% Ad libitum Ad libitum (weeks) 24 g/d 14.5 g/d 17 g/d 11 g/d Body weight 20 491 409 (83%) 276 (56%) 272 239 (88%) 180 (66%) 33 614 482 (78%) 306 (50%) 346 260 (75%) 184 (53%) 60 751 587 (78%) 353 (47%) 412 295 (72%) 212 (51%) 113 712a 597 (84%) 345 (48%) 603 324 (54%) 215 (36%)

Fat content 20 17.4 16.4 9.5 19.9 12.4 6.4 33 24.6 18.4 9.1 24.4 12.2 4.0 60 34.7 26.7 8.6 32.7 14.7 4.8 113 36.5 27.0 8.1 42.2 13.2 5.7 Values derived from reference 27, Tables 2 and 3 A. Rats started the study at 7 wk of age, when males weighed 172 to 266 g and females weighed 134 to 213 g. They were euthanized 13, 26, 53, or 106 wk later. Throughout the study, rats were fed Purina Chow either ad libitum or once daily with the amounts indicated (in g/d and as % of the ad libitum group at the start of the study). Mean body weights (in g, and as % of that of corresponding ad libitum group) have been rounded to the nearest integer. aMean weight loss at 113 weeks may be an artifact of high mortality. press but only after an average interval. This scenario yields con- striction.40 However, although many animals do readily consume stant rates of responding, but because the actual delivery intervals treats, such as cereal or candy, the evidence that they are highly vary unpredictably around the mean, the animals cannot predict motivated to do that is sparse. For example, using a progressive the exact time until the next food reinforcer. When this behavior ratio schedule in which each pellet costs more than the previous had stabilized, rats were tested at various times (1 to 23 h) after pellet, nondeprived rats worked an average of 105 presses for a the previous meal of dog chow. The results, redrawn in Figure total of 6 sweet food pellets whereas rats deprived of food for 24-h 5, show that for equivalent elapsed time since the last meal, rats worked an average of 600 presses for 15 pellets.60 It follows that responded at higher rates on 1-min compared with 3-min vari- rats would not sustain operant performance through a standard able-interval schedules. In addition, response rate increased non- session using palatable treats without some level of restriction linearly with duration since last fed, but fractional changes were or weight loss. Another functional difference between restric- identical at the 2 ratios.10 Important for the present purposes, the tion and nonrestriction protocols was reported by Barbano and rate after a 23-h delay was about double the rate after a 3-h delay. Cador,3 who found that food-restricted rats developed the well- Therefore, if the relative rate of responding on a variable-interval known behavior of heightened activity in anticipation of feeding, schedule can be interpreted as a psychologic manifestation of but nondeprived (palatable food-fed) rats did not. hunger, then animals are about twice as hungry after 23 h as after Given that no completely uniform or agreed-upon levels of re- 3 h without food. This proposition may seem controversial, but it striction exist for specific tasks, investigators tend to use a level of is objectively based. restriction that is more than adequate to ensure that the animals Corroborating data were obtained by Reese and Hogenson,49 will reliably perform the tasks required to answer the scientific who examined pigeons pecking an illuminated key under a fixed- questions. Therefore, even if the minimum amount of depriva- ratio schedule in which each peck produced a 4-s access to a tray tion needed to sustain a task were known for a range of tasks, of grain. At each session, pigeons were allowed to feed until satia- maintaining subjects at that minimum would make performance tion, which was defined as no responses for 30 min. The experi- less reliable, data would inevitably be wasted, and more animal- mental variables were time of deprivation and associated weight days would be necessary to complete the study. Initial weight loss at the time of the test for satiation. Their first result was that restrictions to 85% of free-feeding, followed by an increment ap- the rate of key pecking, and presumably eating, was almost always propriate for the species, sex, and strain, likely will be adequate all-or-none; whenever pigeons were eating, median response rate to ensure strong performance in most behavioral tasks and still be was 4/min. Therefore, pigeons eating a uniform food show rela- consistent with an animal that is at least as healthy as ad libitum- tively abrupt satiation as manifested by transition from uniform fed (and overweight) counterparts. With justification, even lower responding to no responding; this situation is also the case in rats target weights may be acceptable. freely eating from a jar of food.31 However, pigeons did not re- To conclude this section: animals on food restriction regimens spond at all until they were food-deprived for approximately 35 h must be given food that is sufficiently high in quality to ensure or were approximately 85% of the free-feeding weight.49 Most sub- they do not become deficient in any specific nutrient, such as pro- sequent studies in pigeons have used 85% weight as a standard tein or micronutrients. The lower the target weight or food ration, ceiling body weight, with a level of food restriction consistent with the more critical this consideration becomes. If treats are used obtaining this weight, thereby maintaining consistent behavioral as supplements or reinforcers, they should contain protein and performance. Studies in rodents have used comparable levels of micronutrients. weight loss to obtain stable performance. Recommendations regarding food deprivation or restriction. 1) It has been recommended that animals should work for palat- Most species are physiologically equipped to tolerate acute with- able food treats in this type of study, as an alternative to food re- holding of food for periods of as long as 24 h without notewor- 158 Food and fluid restriction

presumably minimize the additional physiologic or psychologic stress of deprivation. Animals have endogenous nycthemeral rhythms that make them particularly adaptable to once-daily occurrences, such as food or water access. Therefore, deprivations of 24 h and once-per-day restricted-access schedules seem to be minimally stressful by the parameters of normal behavior and appearance. Longer periods of acute deprivation or chronic restriction may be acceptable procedures, but ACUCs must ask investigators to implement suitable monitoring protocols, such as routine weighing and target weights. In the case of chronic food restriction, using a fraction of the weight of age-matched free-fed animals as a target may be inappropriate because ad libitum-fed animals become obese and have shortened life expectancy. Spe- cies, age and strain-specific target growth rates may be more appropriate but are not always available. Data from rats have been presented in this review because they are available, but empirical normative data from other species, especially mice, are lacking.

References 1. Armstrong S, Coleman G, Singer G. 1980. Food and water deprivation: changes in rat feeding, drinking, activity and body weight. Figure 5. Responding of rats for food reinforcement during the early Neurosci Biobehav Rev 4:377–402. portion of variable-interval (VI) sessions. Only 1- and 3-min schedules 2. Bachmanov AA, Reed DR, Beauchamp GK, Tordoff MG. 2002. are shown for simplicity; 2-min data were intermediate. The response Food intake, water intake, and drinking spout side preference of 28 rate was higher overall on the shorter-interval schedule, but the degree mouse strains. Behav Genet 32:435–443. changes with time, because the last meals were similar in the 2 conditions. 3. Barbano MF, Cador M. 2005. Various aspects of feeding behavior Figure is redrawn from data in reference 10. can be partially dissociated in the rat by the incentive properties of food and the physiological state. Behav Neurosci 119:1403–1405. thy signs of physiologic or apparent psychologic distress. 2) Food 4. Bartness TJ, Day DE. 2003. Food hoarding: a quintessentially deprivation in excess of 24 h engages a number of physiologic anticipatory appetitive behavior. In: Fluharty SJ, Grill HJ, editors. strategies for saving energy; therefore weight loss is a negatively Progress in psychobiology and physiological psychology. New York: accelerated function with days of deprivation. For many purposes, Academic Press. p 69–100. deprivation in excess of 24 h likely is unnecessary. With scientific 5. Bealer SL, Crofton JT, Share L. 1983. Hypothalamic knife cuts alter justification, deprivation of as long as 72 h is acceptable in rats, fluid regulation, vasopressin secretion, and natriuresis during water deprivation. Neuroendocrinology 36:364–370. producing weight loss in an acceptable range (approximately 15%) 6. Bengele HH, Mathias RS, Perkins JH, Alexander EA. 1981. Urinary without lethargy. Shorter times may be more suitable to achieve concentrating defect in the aging rat. Am J Physiol Renal Physiol 240: comparable physiologic or behavioral effects in mice. 3) Because of F147–F150. endogenous diurnal rhythms of physiology and behavior, animals 7. Berne RM, Levy MN. 1998. Physiology, 4th ed. St Louis: Mosby. often accommodate deprivation or restricted access in multiples 8. Bing C, Frankish HM, Pickavance L, Wang Q, Hopkins DF, Stock of 24 h in duration and periodicity particularly well. 4) Animals MJ, Williams G. 1998. Hyperphagia in cold-exposed rats is accom- adapt well to once-daily food restriction schedules. Water should panied by decreased plasma leptin but unchanged hypothalamic be available either continuously or (at a minimum) for the entire NPY. Am J Physiol Regul Integr Comp Physiol 274:R62–R68. 9. Carlton PL. 1961. The interacting effects of deprivation and reinforce- period provided for consumption of the daily food ration. The ment schedule. J Exp Anal Behav 4:379–381. target intake or weight for such restriction will depend in part 10. Clark FC. 1958. The effect of deprivation and frequency of re- on the species and the objective of the study, but 85% is common inforcement on variable-interval responding. J Exp Anal Behav for rats and birds in behavioral studies. Weight reductions much 1:221–228. less than this may not be sufficient to achieve stable behavioral 11. Collier G, Hirsch E, Hamlin PH. 1972. The ecological determinants performance. Mice may require and tolerate comparable target of reinforcement in the rat. Physiol Behav 9:705–716. weights, although extensive data are not available. 5) Using palat- 12. DeSimone R, Olson C, Erickson R. 1992. The controlled water access able treats instead of deprivation often does not produce sufficient paradigm. ILAR J 34:27–29. 13. Eckel LA. 2004. Estradiol: a rhythmic, inhibitory, indirect control of motivation for behavioral studies. Although treats may be used meal size. Physiol Behav 82:35–41. for environmental enrichment, they should be part of the overall 14. Eckel LA, Moore SR. 2004. Diet-induced hyperphagia in the rat is nutritional program for the animals and the study. influenced by sex and exercise. Am J Physiol Regul Integr Comp Physiol 287:R1080–R1085. 15. Fitzsimons JT. 1979. The physiology of thirst and sodium appetite. Summary and Conclusions Cambridge (UK): Cambridge University Press. Because food and fluid deprivation or restriction has occurred 16. Fox EA, Byerly MS. 2004. A mechanism underlying mature-onset in the evolutionary past of all extant terrestrial species, animals are obesity: evidence from the hyperphagic phenotype of brain-derived physiologically well equipped to deal with such privations. We neurotrophic factor mutants. Am J Physiol Regul Integr Comp have reviewed evidence that shows that after 12 to 24 h without Physiol 286:R994–R1004. access, animals efficiently reduce further fluid or energy losses by 17. Friedman MI, Stricker, EM. 1976. The physiological psychology of a combination of behavioral and physiologic adjustments. These hunger: a physiological perspective. Psychol Rev 83:409–431. 159 Vol 57, No 2 Comparative Medicine April 2007

18. Gannon KS, Smith JC, Henderson R, Hendrick P. 1992. A system for 44. Petersen S, McCarthy JC. 1981. Correlated changes in feeding studying the microstructure of ingestive behavior in mice. Physiol behavior on selection for large and small body size in mice. Behav Behav 51:515–521. Genet 11:57–64. 19. Good DJ. 2005. Using obese mouse models in research: special 45. Pond CM. 1987. Some conceptual and comparative aspects of body considerations for IACUC members, animal care technicians, and composition analysis. In: Toates FM, Rowland NE, editors. Feeding researchers. Lab Animal (NY) 34:30–37. and drinking. Amsterdam: Elsevier. p 499–529. 20. Harris RBS. 1990. Role of set-point theory in regulation of body 46. Purina Mills Inc [Internet]. 2006. Rodent lab diet #5001 [cited June weight. FASEB J 4:3310–3318. 2006]. Available at http://www.labdiet.com. 21. Homma S, Takeda S, Kusano E, Matsuo Y, Shimizu T, Nakamura 47. Radford EP. 1959. Factors modifying water metabolism in rats fed M, Oohara T, Makino S, Asano, Y. 2002. Impaired urinary concen- dry diets. Am J Physiol 196:1098–1108. trating ability in genetially polyuric mice. Nephron 92:889–897. 48. Ramsay DJ, Rolls BJ, Wood RJ. 1977. Body fluid changes which in- 22. Houston AI, McNamara JM. 1989. The value of food: effects of open fluence drinking in the water deprived rat. J Physiol266: 453–469. and closed economies. Anim Behav 37:546–562. 49. Reese TW, Hogenson MJ. 1962. Food satiation in the pigeon. J Exp 23. Hughes JE, Amyx H, Howard JL, Nanry KP, Pollard GT. 1994. Anal Behav 5:239–245. Health effects of water restriction to motivate lever-pressing in rats. 50. Rinaman L, Vollmer RR, Karam J, Phillips D, Li X, Amico JA. 2005. Lab Anim Sci 44:135–140. Dehydation anorexia is attenuated in oxytocin-deficient mice. Am J 24. Jewett JC, Lefever TW, Flashinski DP, Koffarnus MN, Cameron Physiol Regul Integr Comp Physiol 288:R1791–R1799. CR, Hehli DJ, Grace MK, Levine AS. 2006. Intraparaventricular 51. Robertson KL, Rowland NE. 2005. Effect of two types of environ- neuropeptide Y and ghrelin induce learned behaviors that report mental enrichment for singly housed mice on food intake and weight food deprivation in rats. Neuroreport 17:733–737. gain. Lab Anim 34:29–32. 25. Kakolewski JW, Deaux E. 1970. Initiation of eating as a function of 52. Rocha MJ, Chen Y, Oliveira GR, Morris M. 2005. Physiological ingestion of hypo-osmotic solutions. Am J Physiol 218:590–595. regulation of brain angiotensin receptor mRNA in AT1a deficient 26. Keenan KP, Ballam GC, Haught DG, Laroque P. 2000. Nutrition. mice. Exp Neurol 195:229–235. In: Krinke GJ, editor. The laboratory rat. London: Academic Press. 53. Rolls BJ, Rolls ET. 1982. Thirst. Cambridge (UK): Cambridge Uni- p 57–75. versity Press. 27. Keenan KP, Hoe CM, Mixson L, McCoy CL, Coleman JB, Mattson 54. Rowland NE. 1976. Circadian rhythms and partial recovery of regu- BA, Ballham GA, Gumprecht LA, Soper KA. 2005. Diabesity: a latory drinking in rats after lateral hypothalamic lesions. J Comp polygenic model of dietary-induced obesity from ad libitum over- Physiol Psychol 90:382–393. feeding of Sprague-Dawley rats and its modulation by moderate 55. Rowland NE. 1984. Metabolic fuel homeostasis in golden hamsters: and marked dietary restriction. Toxicol Pathol 33:650–674. effects of fasting, refeeding, glucose, and insulin. Am J Physiol Regul 28. Kissileff HR. 1969. Food-associated drinking in the rat. J Comp Integr Comp Physiol 247:R57–R62. Physiol Psychol 67:284–300. 56. Rowland NE. 2002. Thirst and water-salt appetite. In: Pashler H, 29. Kleiber M. 1975. Fire of life. New York: John Wiley & Sons. Gallistel R, editors. Stevens’ handbook of experimental psychology, 30. Le Magnen J. 1992. Neurobiology of feeding and nutrition. San 3rd ed, vol 3. New York: John Wiley & Sons. p 669–707. Diego: Academic Press. 57. Rowland NE, Cansler K, Kim E, Pawlik N, Robertson K. 2002. Fla- 31. Luke RG. 1973. Natriuresis and chloruresis during hydropenia in vor avoidance induced by LiCl and dexfenfluramine in rats and mice the rat. Am J Physiol 224:13–20. using non-deprivation protocols. Behav Neurosci 116:777–784. 32. McKinley MJ, Denton DA, Nelson JF, Weisinger RS. 1983. Dehy- 58. Severinsen T, Munch IC. 1999. Body core temperature during food dration induces sodium depletion in rats, rabbits, and sheep. Am J restriction in rats. Acta Physiologia Scandinavica 165:299–305. Physiol Regul Integr Comp Physiol 245:R287–R292. 59. Strohmayer AJ, Smith GP. 1987. The meal pattern of genetically 33. Marwine A, Collier G. 1979. The rat at the waterhole. J Comp Physiol obese (ob/ob) mice. Appetite 8:111–123. Psychol 93:391–402. 60. Thorpe AJ, Cleary JP, Levine AS, Kotz CM. 2005. Centrally ad- 34. Metzler CH, Thrasher TN, Keil LC, Ramsay DJ. 1986. Endocrine ministered orexin A increases motivation for sweet pellets in rats. mechanisms regulating sodium excretion during water deprivation Psychopharmacology 182:75–83. in dogs. Am J Physiol Regul Integr Comp Physiol 251:R560–R568. 61. Toth LA, Gardiner TW. 2000. Food and water restriction protocols: 35. Moore-Ede MC, Sulzman FM, Fuller CA. 1982. The clocks that time physiological and behavioral considerations. Contemp Top Lab us. Cambridge (MA): Harvard University Press. Anim Sci 39(6):9–17. 36. Morris M, Means S, Oliverio MI, Coffman TM. 2001. Enhanced cen- 62. Tschop M, Heiman ML. 2001. Rodent obesity models: an overview.

tral response to dehydration in mice lacking angiotensin AT1a recep- Exp Clin Endocrinol Diabetes 109:307–319. tors. Am J Physiol Regul Integr Comp Physiol 280:R1177–R1184. 63. Vaughan CH, Moore MC, Haskell-Luevano C, Rowland NE. 2005. 37. Morris P, Mogenson GJ. 1980. Dissociation of nocturnal feeding and Meal patterns and foraging in melanocortin receptor knockout mice. drinking in the rat. Behav Neural Biol 30:299–311. Physiol Behav 84:129–133. 38. Mrosovsky N, Powley TL. 1977. Set points for body weight and fat. 64. Vaughan CH, Moore MC, Haskell-Luevano C, Rowland NE. 2006. Behav Biol 20:205–223. Food motivated behavior of melanocortin-4 receptor knockout mice 39. Munch IC. 1995. Influences of time intervals between meals and under a progressive ratio schedule. Peptides 27:2829–2835. total food intake on resting metabolic rate in rats. Acta Physiol Scand 65. Vaughan CH, Rowland NE. 2003. Meal patterns of lean and leptin- 153:243–247. deficient obese mice in a simulated foraging environment. Physiol 40. National Research Council. 1996. Guide for the care and use of Behav 79:275–279. laboratory animals. Washington (DC): National Academy Press. 66. Watts AG. 1999. Dehydration-associated anorexia: development and 41. Nelson RJ. 1988. Restricted water intake influences male reproduc- rapid reversal. Physiol Behav 65:871–878. tion in two strains of house mice (Mus musculus). Physiol Behav 67. Watts AG. 2001. Neuropeptides and the integration of motor re- 43:217–221. sponses to dehydration. Annu Rev Neurosci 24:357–384. 42. Paque C. 1980. Sahara Bedouins and the salt water of the Sahara: a 68. Wellman PJ, Bellinger LL, Cepeda-Benito A, Susabda A, Ho DHH, model for salt intake. In: Kare MR, editor. Biological and behavioral Davis KW. 2005. Meal patterns and body weight after nicotine in aspects of salt intake. New York: Academic Press. p 31–47. male rats as a function of chow or high-fat diet. Pharmacol Biochem 43. Peck JW. 1978. Rats defend different body weights depending on Behav 82:627–634. palatability and accessibility of their food. J Comp Physiol Psychol 69. Wirtshafter D, Davis JD. 1977. Set points, settling points, and the 92:555–570. control of body weight. Physiol Behav 19:75–78.

160